Improving Reordering with Linguistically Informed Bilingual n-grams

We present a new reordering model estimated as a standard n-gram language model with units built from morpho-syntactic information of the source and target languages. It can be seen as a model that translates the morpho-syntactic structure of the input sentence, in contrast to standard translation models which take care of the surface word forms. We take advantage from the fact that such units are less sparse than standard translation units to increase the size of bilingual context that is considered during the translation process, thus effectively accounting for mid-range reorderings. Empirical results on French-English and German-English translation tasks show that our model achieves higher translation accuracy levels than those obtained with the widely used lexicalized reordering model.